Case Study - Project iDEA: intelligent Digital-Emotion Analyser Library

The Project

The main idea behind project iDEA is to model the evolution of concepts as a timewave of emotion. It is a search engine that extracts events from search results, thus producing a timeline of the main events associated with whatever that was searched. It then determines the overall decrease or increase in emotion for each event and plots each percentage change relative to time, thereby generating an overall "Emotion Timewave" (ET) for the searched concept.

Project iDEA presents a new way to model the flow of historic and current events, and will hopefully prove insightful into the nature of future events through ET pattern recognition. The accuracy of the textual data numerical analysis is determined by the intelligence of the "Emotional Interpreter" (EI); this module interprets prose by emotionally digitising words and sentences.

There are many solutions to design. How exactly can events be extracted from historic data? How do we decide the degree of negativity or positivity of an emotion? How can we ensure we calculate the intended emotional interpretation of a sentence? How do we distinguish the affective subjects of an event and separate their "Emotional Responses" (ER)? Can we start to design the structure of an "emotions database"? All will be explored in this project.

Initial data sources to extract from will be Wikipedia for historic data and ideally BBC News for current data. The data collected will be used to produce graphs and timelines using the BFO Graph Library.

The project could be limited to analysing current data as it is naturally eventful, e.g. BBC News - data that has an associated date. This would concentrate the project towards the emotion research without the need to worry about event extraction, although it would only represent a relatively small fraction of history (1997 onwards).

Eugene Yuta Bann
Computer Scientist
Bath University